[gmx-users] Using a 4.6.5 cpt file in 5.0 to extend a run?

[Nathan Scott]

Mark

Thank you very much for your help. The one part of your reply that confuses
me a bit is that I *did* actually pass the old cpt file to grompp to
generate a new tpr file before beginning mdrun. I should have been clear
about that in my original post. To be specific, below is the grompp command
I used:

grompp_mpi -f stn_wt_d_in_1.mdp -c stn_wt_c_1.gro -t stn_wt_c_1.cpt -p
../topol_exc.top -o stn_wt_d_1.tpr -po stn_wt_d_out_1.mdp -e stn_wt_c_1.edr

So I pass to grompp my new mdp file, the old (from 4.6.5) gro, cpt, edr,
and topology files, and generate a tpr file based on that information.
Grompp doesn't complain in any way, and yet mdrun does.

Your point about switching versions mid-project is very well taken, and
that was my first thought as well. I've got an email in to XSEDE help to
address the issue, but I'm not expecting much via that route.

Best Wishes,
Nathan

On Thu, Sep 25, 2014 at 3:55 AM, Mark Abraham <mark.j.abraham at gmail.com>
wrote:

> On Thu, Sep 25, 2014 at 5:03 AM, Nathan Scott <scottjn at gmail.com> wrote:
>
> > Hi all,
> >
> > I am trying to extend some simulations done on the XSEDE Trestles system
> > over the summer, and they have upgraded to 5.0 since then from 4.6.5.
> >
>
> That's not normal practice at compute centers - older versions should
> generally continue to be made available, so that a scientific project can
> have a consistent methodology. In principle, a trajectory could be
> continued with newer software if it implements the same method, but proving
> that this principle holds is more trouble than you should normally want to
> buy...
>
> Can anyone give any tips for navigating around the below error?
> >
> > Program mdrun_mpi, VERSION 5.0
> > Source code file:
> >
> >
> /state/partition1/git/chemistry-roll/BUILD/gromacs_intel_mvapich2_ib-5.0/gromacs-5.0/src/gromacs/gmxlib/checkpoint.c,
> > line: 2037
> >
> > Fatal error:
> > You seem to have switched ensemble, integrator, T and/or P-coupling
> > algorithm between the cpt and tpr file. The recommended way of doing this
> > is passing the cpt file to grompp (with option -t) instead of to mdrun.
> If
> > you know what you are doing, you can override this error by setting the
> > env.var. GMX_ALLOW_CPT_MISMATCH
> > For more information and tips for troubleshooting, please check the
> GROMACS
> > website at http://www.gromacs.org/Documentation/Errors
> >
> > In fact, I haven't switched any of the mdp options that were used for the
> > original simulation, except nsteps, and doing a diff of the mdp file the
> > original cpt file is based off of and the new one that I am trying to use
> > to extend the simulation confirms this fact.
> >
>
> OK, but sometimes the internal representation has to change, as it did for
> (say) the new implementation of the default sd integrator, or the default
> for cutoff-scheme. Depending what the actual issue is here, maybe the error
> message could use improvement.
>
> Is there a workaround for this?
>
>
> Like the message says, you can do
>  grompp -f new -c whatever -p topol -t the-4.6.5.cpt
>
>  in order that your intentions are explicit. mdrun should have logic to
> stop you doing things that likely are mistakes, such as
>
>  mdrun -s nvt -cpi npt
>
> but we can't write code that anticipates all possible scenarios and reacts
> the way every user would expect...
>
> Or might I just be making some kind of
> > simple mistake? Any advice would be greatly appreciated!
> >
> > Best Wishes,
> > Nathan
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> > http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> > posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> > send a mail to gmx-users-request at gromacs.org.
> >
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>